Neural Network Modeling of the Strength of Normal Sections of Prefabricated Reinforced Concrete Ribbed Slabs

Introduction. Precast reinforced concrete ribbed slabs are broadly used as floors and coverings for industrial, residential and public buildings. Their use in this capacity is due to the high technological efficiency of manufacturing, efficient use of concrete and the possibility of automating factory production. One of the critical tasks in designing such structures is to calculate the bearing capacity of normal cross sections. Traditional calculation methods are reliable, but they are outdated. Machine learning methods are increasingly being employed in engineering, where researchers are opting for artificial neural networks (ANNs). The use of traditional methods in processing structured data such as tables and databases has its limitations. Neural networks are capable of analyzing unstructured data such as text, images, and videos, which opens up new prospects for analyzing and comprehending information. The article sets forth an approach to neural network modeling of the bearing capacity of normal sections of prefabricated reinforced concrete ribbed slabs. Materials and Methods. A structured and processed data array (dataset) includes 20 samples for which a computational model based on a multilayer perceptron has been developed and verified. The input parameters are the geometric as well as physical and mechanical characteristics of the slabs and the applied load, the output parameter is the limiting bending moment calculated using the limit state method. Research Results . Training on a limited sample did not lead to retraining of the model due to the correct division of data into test, training and control batches and the use of the quasi-Newton optimization method. The model has displayed a high level accuracy and reliability. Artificial neural networks are capable of identifying nonlinear dependencies between the parameters with no a priori assumptions. Discussion and Conclusion. The suggested model is not a substitute for the existing calculations, but it serves as an efficient digital tool for quick verification of design solutions, optimization of reinforcement and improvement of structural reliability. Its implementation into BIM systems and digital construction platforms is in compliance with the requirements of Industry 4.0 and creates new opportunities for designing prefabricated reinforced concrete structures.